Hybrid approach to bring intelligence and resilience in capital market
As we all aware of the significance of data in our life, we can safely refer this era more of a data driven age that has the potential of fusing enormous volume of data from disparate sources. How data helps us to get deeper insights about the new dimensions in analytics that is an open item for discussions in research seminars. However, if we try to appreciate how data is changing each industry, we will see there is a common pattern. But before we delve in there to acknowledge the proficiency of data driven analytics and cosmic power of process mining capabilities, we need to understand the high-level models and its interdependence both from cognitive and interactive perspectives.
We know data driven systems can design real and tangible evidence to improve transactional protocols in capital market. Most of the methods used in analyzing capital market phenomenon are descriptive statistics which has a foundation on classical methodologies. Researchers are focusing more on identifying the fine line between knowledge and certainty. Knowledge is merely a repository of different scenarios and its outcomes, now the question is whether this collection of scenarios is inclusive? Answer is no, then what enables us to think of certainty if there is gap in understanding different dimensions comprehensively. That’s where data aids us to get the insights by extrapolating the observed values.
To put something in perspective certainty is a measure of confidence that can be aligned to an event. In this regard we all know the importance of value at risk (VaR), however it still has some limitations as VaR is determined as a function of mean and variance of the returns and it is surmising normal distribution. In a typical contextual model knowledge is what best we know from fundamental analysis or technical analysis, it’s the best possible assimilation of all validated sources and certainty is the technique that we will deploy to map the indicators of the future events to group or classify those into specific category.
We can also infer certainty as the model that will help us to derive the future knowledge. Cognitive computing will enable us to transpose human interventions to mere machine capabilities. However, if we think rationally, we are reducing the confidence level as machines are less powerful to support innovations and process unique diegesis.
领英推荐
This could be a probable frailty of exhaustive cognitive approach. We can resolve this issue by embracing a mid-way, human centric cognitive approach that is precisely based on interactive model, that augments the interactions between human and machine, it is just trying to replicate human brain as another computational intersection. Machine may add value using unrivalled memory power and impeccable computational capabilities that is required to determine pricing of securities accurately, however human brain can accentuate the interactions using heuristics, like model selection on beta regression for machine learning systems, which needs human supervision. Monte Carlo simulation that is predominantly used for complex asset pricing, results on the finite sample behavior of prediction-based model. That’s how the efficiency of future markets may be enhanced through a hybrid approach. Some researchers have already demonstrated that if we try to explore best price for securities in capital market, human brain and machine interactions has its combined efficacy in reducing the convergence time compared to classical cognitive approach. However further research is required to quantify the benefits and substantiate the value propositions of this hybrid approach.
Thoughts, comments are welcome!
Assistant Professor, Madras School of Economics | PhD in Economics, Indian Institute of Management Kozhikode | Recipient of Reserve Bank of India Faculty Scholarship - 2024
3 年Great read Sir!! As you have rightly pointed out "embracing the mid way" between human and machine interaction is a great beginning and the utility lies in almost all areas dealing with enormous amount of data.
Debt Advisory I Financial Services I Climate Finance| Doctoral Scholar @XLRI
3 年Well thought of article. I have some thoughts. Uncertainty is the key to capital markets. So when you are talking of merging human interaction with machine learning then i feel the key is to understand the predictability of human brain. I deal with these people day in day out and i find that no proper predictions or justifications can be given for the behavior. At times its information at times its influence at times its whims. Challenge is ro predict the decisive factor. When we talk of information role of media is big. But to quantify and transform it to prediction is the chlange
Process Engineer at Wells Fargo
3 年Good one
Business Unit Executive, IBM Z Technical Sales, Americas
3 年Raja - there are definitely some very interesting possibilities. Clearly insight is more than what a computer (of any type) can create because people are involved. Your predictive model will need to integrate the uncertainty of psychology into an analytic model that combines a large variety of possibilities to generate a scoring mechanism that can predict stock prices. We too often look at people as acting as a system would - but that is not the whole story. Things like weather, empathy, social conditions, perspective on companies based on country relationships, current events, pending legislation or import concerns, or simply the fact that people follow other people (sometimes blindly) can have significant impact behaviors that will impact the pricing. It is definitely going to be a long and interesting ride........just don't assume that it is always fully rational or explainable because people are involved.
Global Research Leader in Banking, IBM Institute for Business Value | Bestselling author | Podcaster | Board advisor | International speaker
3 年The key issue is that “fundamental uncertainty” is the structure of financial markets. Also AI must face fundamental uncertainty to avoid collapse of the algorithm because, as in quantum theory, reality cannot be fully described by information. As such, in risk management the value of VaR is not about “what is measured” but enabling a discussion about “what could not be measured” but must be considered. There is the space (uncertainty) in which investors make significant money or go bust. Therefore, only learning how to use AI to help opening the reference framework of the decision-making truly adds value. What matters, however, is that the scope is not “stability” anymore (which cannot be reached in the presence of information gaps which also AI faces on big data), but “anti fragility”. In this case, the process of decision making is where you make the opening. This leads to a usage of AI outside the mathematical optimization, inside the what-if generator. You will enjoy new book “Banks and Fintech on Platform Economies” that links all these topics.